As the overlay network of the existing Internet, Peer to Peer (P2P) network shares the computer resources and services with the peer to peer exchanging between the systems to achieve distributed, reliable, extendable and robust network application mode. It breaks through the conventional client/server (C/S) mode, and all nodes are equivalent in the network. Each node works as the server to provide services to other nodes and also enjoys the services provided by other nodes.
With the rapid development of Internet and broadband access network, P2P technology has a lot of applications and research, especially, the whole distributed and structured P2P network (that is, the DHT network) based on the distributed hash table (DHT) and the overlay network is widely paid attention in the industry.
In the DHT network, some nodes might exit the network normally or abnormally, which easily leads to the lose of the storage resource, in order to enhance the availability of the storage resource, that is, to avoid a situation in which the resource cannot be acquired due to the failure of an individual node, some other nodes are generally used to back up the resource in the DHT network, and the node which is used to back up a certain resource is called the backup node of the resource. Backing up resource improves the availability of the resource.
In the prior art, the resource is mainly backed up internally in one DHT network. For example, in the Chord, the nodes are arranged into a DHT ring clockwise from small to large according to the node ID, a certain node takes the first node encountered in the clockwise direction as its backward node, and since each node maintains its backward node list, the data stored in the node are directly backed up in its backward node, and if the data should be backed up for several copies, the data should be saved in multiple backward nodes. The maintenance of this backup method is relatively simple, but in some abnormal situations, however, this method cannot guarantee the availability of the resource. There are a lot of methods in the prior art to improve the backup method internal in the DHT network, and the improvements are mainly focused on two sides: one is backing up the data for several copies and storing the copies in a plurality of nodes according to certain rules, and only taking out one copy in recovery; the other is dividing the data into several redundant segments, therefore, only some segments are needed to recover the data, so as to achieve backing up the data, the storage amount of this backup method is relatively small, but the maintenance overhead is relatively big.
Since the abovementioned backup method only works within the DHT network, when there are few nodes exiting the network, this method can recover the network data very well, but when there are lots of nodes exiting abnormally, some data might cannot be recovered, which is not permitted by networks with very high reliability requirements.